Reference
Reference: server.ai
server.ai.allow_remote
| Syntax | allow_remote: true | false |
| Default | false |
| Context | server.ai |
| Reload | needs a restart |
Allow a non-loopback endpoint (telemetry leaves the host).
server.ai.api_key_env
| Syntax | api_key_env: <string> |
| Default | none |
| Context | server.ai |
| Reload | needs a restart |
Environment variable holding an API key, if the endpoint needs one.
server.ai.ca_bundle
| Syntax | ca_bundle: <string> |
| Default | none |
| Context | server.ai |
| Reload | needs a restart |
Extra CA (PEM) trusted for an https:// endpoint, besides the system roots.
server.ai.enabled
| Syntax | enabled: true | false |
| Default | false |
| Context | server.ai |
| Reload | needs a restart |
Turn the AI advisor on.
server.ai.endpoint
| Syntax | endpoint: <string> |
| Default | http://127.0.0.1:8080/v1/chat/completions |
| Context | server.ai |
| Reload | needs a restart |
OpenAI-compatible chat-completions URL, http:// or https:// (ca_bundle for a private CA).
server.ai.max_queue
| Syntax | max_queue: <integer> |
| Default | 4 |
| Context | server.ai |
| Reload | needs a restart |
Pending requests; more are refused.
server.ai.max_tokens
| Syntax | max_tokens: <integer> |
| Default | 800 |
| Context | server.ai |
| Reload | needs a restart |
Maximum length of one answer, in tokens.
server.ai.min_interval
| Syntax | min_interval: <duration> (e.g. 30s) |
| Default | 600s |
| Context | server.ai |
| Reload | needs a restart |
Minimum time between automatic requests for one application.
server.ai.model
| Syntax | model: <string> |
| Default | “ |
| Context | server.ai |
| Reload | needs a restart |
Model name as the endpoint knows it; chosen by the operator.
server.ai.timeout
| Syntax | timeout: <duration> (e.g. 30s) |
| Default | 60s |
| Context | server.ai |
| Reload | needs a restart |
Time limit for one request to the model endpoint.